Question answering device and question answering method

Abstract

A question answering device (2) is configured so as to comprise: a sensing data collection unit (21) that collects sensing data of a sensor for monitoring the situation of a person to be cared; and a question data acquisition unit (22) that acquires question data indicating a question related to the person to be cared. The question answering device (2) also comprises: an answer creation unit (23) that, on the basis of the sensing data collected by the sensing data collection unit (21), creates an answer to the question indicated by the question data acquired by the question data acquisition unit (22); and an answer data reply unit (24) that replies answer data indicating the answer created by the answer creation unit (23).

Description

Question and answering device and question and answering method

[0001] The present disclosure relates to a question and answering device and a question and answering method.

[0002] There is a question answering device that generates answers to questions about care recipients. For example, Patent Literature 1 discloses a question answering device that includes an acquisition unit, a calculation unit, and an extraction unit. The acquisition unit acquires a query sentence indicating the content of a question from a user terminal. The calculation unit calculates the similarity between each of a plurality of answer candidate sentences that describe an answer to the question and the query sentence. The extraction unit extracts, from the plurality of answer candidate sentences, an answer candidate sentence that has a relatively high similarity calculated by the calculation unit, and transmits the extracted answer candidate sentence to the user terminal.

[0003] Japanese Patent Application Laid-Open No. 2021-124824

[0004] The answer candidate sentences extracted by the extraction unit of the question answering device disclosed in Patent Document 1 are answer candidate sentences with a relatively high degree of similarity among multiple answer candidate sentences, but there is a problem in that they do not necessarily indicate an answer that reflects the situation of a specific person related to the question.

[0005] The present disclosure has been made to solve the above-mentioned problems, and aims to provide a question answering device that can create answers to questions that reflect the situation of the person being cared for.

[0006] A question answering device according to the present disclosure includes a sensing data collection unit that collects sensing data from sensors that monitor the condition of a care recipient, and a question data acquisition unit that acquires question data indicating a question about the care recipient. The question answering device also includes an answer creation unit that creates an answer to a question indicated by the question data acquisition unit based on the sensing data collected by the sensing data collection unit, and an answer data reply unit that replies with answer data indicating the answer created by the answer creation unit.

[0007] According to the present disclosure, it is possible to create an answer to a question that reflects the condition of the care recipient.

[0008] 1 is a configuration diagram showing a system including a question and answering apparatus 2 according to a first embodiment. FIG. 2 is a hardware configuration diagram showing the hardware of the question and answering apparatus 2 according to the first embodiment. FIG. 3 is a hardware configuration diagram of a computer when the question and answering apparatus 2 is realized by software, firmware, or the like. FIG. 4 is a flowchart showing a question and answering method, which is a processing procedure of the question and answering apparatus 2. FIG. 5 is an explanatory diagram showing the answer "You need ◇◇ ml more" when the question is "How much more water does XX need?". FIG. 6 is a configuration diagram showing a system including a question and answering apparatus 2 according to a second embodiment. FIG. 7 is a hardware configuration diagram showing the hardware of the question and answering apparatus 2 according to the second embodiment. FIG. 8 is an explanatory diagram showing the answer "You need ◇◇ ml more" when the question is "How much more water does XX need?", advice "Because XX has diabetes, there is a target amount of water intake. Please encourage XX to drink more," and a graph showing changes in sensing data over time. FIG. 9 is a configuration diagram showing a system including a question and answering apparatus 2 according to a third embodiment. FIG. 10 is a hardware configuration diagram showing the hardware of the question and answering apparatus 2 according to the third embodiment.

[0009] In order to explain the present disclosure in more detail, embodiments of the present disclosure will be described below with reference to the accompanying drawings.

[0010] Embodiment 1. Fig. 1 is a configuration diagram showing a system including a question and answering apparatus 2 according to embodiment 1. Fig. 2 is a hardware configuration diagram showing the hardware of the question and answering apparatus 2 according to embodiment 1. The system shown in Fig. 1 includes a chatbot 1, a question and answering apparatus 2, and a sensing device 3.

[0011] The chatbot 1 includes a voice recognition unit 11, a voice database 12, a chatbot engine unit 13, a display unit 14, and a voice output unit 15. The chatbot 1 recognizes the user's voice and outputs text data indicating the user's voice to the question answering device 2. In the system shown in FIG. 1 , the text data indicates a question about the care recipient, and the text data is output as question data to the question answering device 2. The chatbot 1 outputs the answer data returned from the question answering device 2 to the display unit 14 and the voice output unit 15, respectively.

[0012] The speech recognition unit 11 recognizes the user's speech and creates text data representing the user's speech based on the speech data recorded in the speech database 12. The speech data recorded in the speech database 12 is, for example, speech piece data representing speech pieces. The speech recognition unit 11 outputs the text data as question data to the chatbot engine unit 13. The speech database 12 stores, for example, speech piece data and text data corresponding to the speech piece data.

[0013] When the chatbot engine unit 13 acquires question data from the voice recognition unit 11, it outputs the question data to the question answering apparatus 2, and when text data is provided as question data from a keyboard or the like (not shown), it outputs the text data to the question answering apparatus 2. The chatbot engine unit 13 acquires answer data from the question answering apparatus 2. The chatbot engine unit 13 displays the answer indicated by the answer data on the display unit 14. The chatbot engine unit 13 outputs the answer indicated by the answer data by voice from the audio output unit 15.

[0014] The display unit 14 displays on a display (not shown) the answer indicated by the answer data output from the chatbot engine unit 13. The audio output unit 15 outputs the answer indicated by the answer data output from the chatbot engine unit 13 as audio from a speaker (not shown).

[0015] The question and answer device 2 includes a sensing data collection unit 21, a question data acquisition unit 22, an answer creation unit 23, and an answer data reply unit 24. The sensing data collection unit 21 is realized, for example, by a sensing data collection circuit 31 shown in FIG. 2. The sensing data collection unit 21 collects sensing data from various sensors included in the sensing device 3. The various sensors are sensors that monitor the condition of the care recipient. The condition of the care recipient may, for example, be the vital signs of the care recipient, the behavior of the care recipient, or the situation in which the care recipient is placed. The situation of the care recipient may, for example, be the temperature, humidity, or ambient noise in the care recipient's room. The sensing data collection unit 21 outputs the sensing data to the answer creation unit 23.

[0016] The question data acquisition unit 22 is realized, for example, by the question data acquisition circuit 32 shown in Fig. 2. The question data acquisition unit 22 acquires question data indicating a question about the care recipient from the chatbot engine unit 13 of the chatbot 1. The question data acquisition unit 22 outputs the question data to each of the sensing data collection unit 21 and the answer creation unit 23.

[0017] The answer creation unit 23 is realized, for example, by the answer creation circuit 33 shown in Fig. 2. The answer creation unit 23 acquires sensing data from the sensing data collection unit 21 and acquires question data from the question data acquisition unit 22. The answer creation unit 23 creates an answer to a question indicated by the question data based on the sensing data and the care record of the care recipient. The answer creation unit 23 outputs answer data indicating the created answer to the answer data return unit 24.

[0018] Specifically, the answer creation unit 23 provides the question data to the learning model 41 and acquires, from the learning model 41, a template answer to the question indicated by the question data. The answer creation unit 23 acquires the care record of the care recipient from the care record database 42. The answer creation unit 23 creates an answer to the question indicated by the question data based on the template answer, the sensing data, and the care record of the care recipient. The learning model 41 is realized, for example, by a neural network. During learning, the learning model 41 is provided with question data indicating a question about the care recipient and template data indicating a template answer corresponding to the question, and learns the template answer. The template data is training data. During inference, when question data is provided from the answer creation unit 23, the learning model 41 outputs template data indicating a template answer corresponding to the question indicated by the question data. The care record database 42 stores the care records of the care recipient. In the system shown in FIG. 1 , the learning model 41 and the care record database 42 are both provided outside the question answering device 2. However, this is merely an example, and the learning model 41 and the care record database 42 may each be provided inside the question and answering device 2.

[0019] The answer data reply unit 24 is realized by, for example, the answer data reply circuit 34 shown in Figure 2. The answer data reply unit 24 acquires answer data from the answer creation unit 23. The answer data reply unit 24 replies with the answer data to the chatbot engine unit 13 of the chatbot 1.

[0020] 1, it is assumed that each of the components of the question and answering apparatus 2, namely, the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 23, and the answer data reply unit 24, is realized by dedicated hardware as shown in Fig. 2. That is, it is assumed that the question and answering apparatus 2 is realized by a sensing data collection circuit 31, a question data acquisition circuit 32, an answer creation circuit 33, and an answer data reply circuit 34. Each of the sensing data collection circuit 31, the question data acquisition circuit 32, the answer creation circuit 33, and the answer data reply circuit 34 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0021] The components of the question answering apparatus 2 are not limited to those realized by dedicated hardware, and the question answering apparatus 2 may be realized by software, firmware, or a combination of software and firmware. Software or firmware is stored as a program in the memory of a computer. A computer refers to hardware that executes a program, and includes, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a central processing unit, a processing unit, an arithmetic unit, a microprocessor, a microcomputer, a processor, or a DSP (Digital Signal Processor).

[0022] 3 is a hardware configuration diagram of a computer when the question and answering apparatus 2 is realized by software, firmware, etc. When the question and answering apparatus 2 is realized by software, firmware, etc., programs for causing a computer to execute the respective processing procedures of the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 23, and the answer data reply unit 24 are stored in memory 51. Then, a processor 52 of the computer executes the programs stored in memory 51.

[0023] 2 shows an example in which each of the components of the question answering apparatus 2 is realized by dedicated hardware, and Fig. 3 shows an example in which the question answering apparatus 2 is realized by software, firmware, etc. However, this is merely an example, and some of the components in the question answering apparatus 2 may be realized by dedicated hardware and the remaining components may be realized by software, firmware, etc.

[0024] The sensing device 3 includes sensors for monitoring the condition of the care recipient, such as a bed sensor 61, a room sensor 62, an appliance sensor 63, a human presence sensor 64, a vital sign sensor 65, and an ingestion detection sensor 66, and also includes a sensing data recording unit 67. The bed sensor 61 is a sensor attached to the bed used by the care recipient and monitors, for example, the state of the care recipient turning over in bed. The room sensor 62 is a sensor installed in the care recipient's room and monitors, for example, the care recipient's sleep, falls, or body movements.

[0025] The home appliance sensor 63 is, for example, a sensor attached to an air conditioner in the care recipient's room and monitors, for example, the room temperature in the room. The home appliance sensor 63 is also a sensor attached to a home appliance such as a refrigerator, stove, rice cooker, or microwave and monitors, for example, the usage status of the home appliance. The home appliance is used by the care recipient. Knowing the usage status of the home appliance makes it possible to determine the care recipient's level of independence. The human presence sensor 64 is, for example, a sensor attached to the entrance / exit of the care recipient's room and detects, for example, when the care recipient leaves the room. The vital sign sensor 65 is, for example, a sensor attached to the wall, ceiling, furniture, or the care recipient's room and monitors, for example, the vital signs of the care recipient. The intake detection sensor 66 is, for example, a sensor equipped with a camera attached to the wall of the care recipient's room and monitors, for example, the care recipient's water intake behavior or food intake behavior. The sensing data recording unit 67 is a database that records the sensing data from the bed sensor 61 , the room sensor 62 , the home appliance sensor 63 , the human presence sensor 64 , the vital sensor 65 and the ingestion detection sensor 66 .

[0026] Next, we will explain the operation of the question answering device 2 shown in Fig. 1. Fig. 4 is a flowchart showing a question answering method, which is a processing procedure of the question answering device 2. In the system shown in Fig. 1, the care recipient is, for example, a person hospitalized in a nursing home or a medical hospital, and the user who asks a question about the care recipient is, for example, a staff member of the nursing home or the medical hospital.

[0027] When the voice of the staff member is provided to the voice recognition unit 11 of the chatbot 1 via a microphone (not shown), the voice recognition unit 11 breaks down the voice of the staff member into, for example, a plurality of speech segments. The voice recognition unit 11 acquires text data corresponding to each speech segment from the voice database 12. The voice recognition unit 11 connects the text data corresponding to each speech segment and outputs the connected text data to the chatbot engine unit 13 as question data.

[0028] The question data may include, for example, "How much more water does Mr. / Ms. XX need to drink?", "Has anything changed about Mr. / Ms. XX?", "Has Mr. / Ms. XX been sleeping well recently?", or "How active was Mr. / Ms. XX during the day yesterday?"

[0029] Here, the staff member's voice is provided to the voice recognition unit 11, and text data is output as question data from the voice recognition unit 11 to the chatbot engine unit 13. However, this is merely an example, and the staff member may operate a keyboard or the like (not shown) to provide text data as question data to the chatbot engine unit 13 from the keyboard or the like.

[0030] The chatbot engine unit 13 acquires question data from the voice recognition unit 11. The chatbot engine unit 13 outputs the question data to the question data acquisition unit 22 of the question and answering apparatus 2.

[0031] The question data acquiring unit 22 acquires question data indicating a question about the care recipient from the chatbot engine unit 13 (step ST1 in FIG. 4 ). The question data acquiring unit 22 outputs the question data to the answer creating unit 23.

[0032] The sensing data collection unit 21 collects sensing data related to the care recipient from the sensing data recorded in the sensing data recording unit 67 of the sensing device 3 (step ST2 in FIG. 4 ). Specifically, the sensing data collection unit 21 identifies the care recipient by performing a care recipient identification process. The care recipient identification process may, for example, be a well-known process of identifying the care recipient by searching for the care recipient's name included in the question data output from the question data acquisition unit 22. The sensing data collection unit 21 also identifies the sensing data to be collected by performing a collection target sensing data identification process. The sensing data identification process may, for example, be a well-known process of identifying the sensing data to be collected by searching for sensing data related to terms included in the question data output from the question data acquisition unit 22.

[0033] For example, if the question is "How much more water does Mr. / Ms. X need?", the sensing data collecting unit 21 identifies sensing data from the ingestion detection sensor 66 related to "water." For example, if the question is "Have you been sleeping well recently, Mr. / Ms. X?", the sensing data collecting unit 21 identifies sensing data from the bed sensor 61 or sensing data from the room sensor 62 related to "sleeping." For example, if the care recipient is Mr. / Ms. X and the sensing data to be collected is sensing data from the ingestion detection sensor 66, the sensing data collecting unit 21 collects sensing data from the ingestion detection sensor 66 related to Mr. / Ms. X. For example, if the care recipient is Mr. / Ms. YY and the sensing data to be collected is sensing data from the bed sensor 61, the sensing data collecting unit 21 collects sensing data from the bed sensor 61 related to Mr. / Ms. YY. Here, the sensing data collecting unit 21 identifies which sensing data is to be collected by searching for sensing data related to terms included in the question data. However, this is merely an example, and the sensing data collection unit 21 may identify which sensing data is to be collected by referring to a database in which sensing data corresponding to terms included in the question data is recorded. The sensing data collection unit 21 outputs the collected sensing data to the answer creation unit 23.

[0034] The answer creation unit 23 acquires sensing data from the sensing data collection unit 21 and acquires question data from the question data acquisition unit 22. The answer creation unit 23 also acquires the care record of the care recipient from the care record database 42. The answer creation unit 23 creates an answer to the question indicated by the question data based on the sensing data and the care record of the care recipient (step ST3 in FIG. 4). The answer creation process by the answer creation unit 23 will be specifically described below.

[0035] The answer creation unit 23 provides the question data to the learning model 41 and acquires, from the learning model 41, an answer template for the question indicated by the question data. If the question indicated by the question data is, for example, "How much more water does Mr. / Ms. XX need?", then, for example, an answer template of "You need ◇◇ ml more" is acquired as answer template (1) and an answer template of "The required amount of water has been acquired" is acquired as answer template (2). If the question indicated by the question data is, for example, "Have you been sleeping well recently?", then, for example, an answer template of "Mr. / Ms. XX has been sleeping well recently" is acquired as answer template (1) and an answer template of "Mr. / Ms. XX has not been sleeping well recently" is acquired as answer template (2).

[0036] For example, if the question indicated by the question data is "How much more water does Mr. / Ms. XX need to drink?", the answer creating unit 23 obtains information on "the amount of water Mr. / Ms. XX needs to drink per day" from Mr. / Ms. XX's care record stored in the care record database 42. If Mr. / Ms. XX's medical history includes diabetes, the information on "the amount of water Mr. / Ms. XX needs to drink per day" may be included in Mr. / Ms. XX's care record. The answer creating unit 23 obtains information on "the amount of water Mr. / Ms. XX has consumed today" from the sensing data of the intake detection sensor 66 recorded in the sensing data recording unit 67. Mr. / Ms. XX's amount of water consumed per unit time is recorded in the sensing data recording unit 67 as time-series data.

[0037] If "XX's daily water intake" is equal to or greater than "XX's required daily water intake," the answer creation unit 23 selects answer template (2) from answer template (1) and answer template (2). Then, the answer creation unit 23 outputs answer template (2) to the answer data reply unit 24 as answer data indicating the answer to the question. If "XX's daily water intake" is less than "XX's required daily water intake," the answer creation unit 23 selects answer template (1) from answer template (1) and answer template (2). Then, the answer creation unit 23 embeds the difference between "XX's required daily water intake" and "XX's daily water intake" in "◇◇" in answer template (1), and outputs answer template (1) with the embedded difference to the answer data reply unit 24 as answer data indicating the answer to the question.

[0038] For example, if the question indicated by the question data is "Has Mr. / Ms. XX been sleeping well recently?", the answer creating unit 23 acquires information on "Ms. XX's average sleeping hours" from the care record of Mr. / Ms. XX stored in the care record database 42. The answer creating unit 23 acquires information on "Ms. XX's sleeping hours yesterday" from the sensing data of the bed sensor 61 or the sensing data of the room sensor 62 recorded in the sensing data recording unit 67.

[0039] If "X's sleeping time yesterday" is equal to or greater than "X's average sleeping time," the answer creation unit 23 selects answer template (1) from answer template (1) and answer template (2). Then, the answer creation unit 23 outputs answer template (1) to the answer data reply unit 24 as answer data indicating an answer corresponding to the question. If "X's sleeping time yesterday" is less than "X's average sleeping time," the answer creation unit 23 selects answer template (2) from answer template (1) and answer template (2). Then, the answer creation unit 23 outputs answer template (2) to the answer data reply unit 24 as answer data indicating an answer corresponding to the question.

[0040] In the question answering device 2 shown in FIG. 1 , the answer creation unit 23 creates an answer to a question indicated by question data based on sensing data and the care record of the care recipient. However, this is merely an example, and the answer creation unit 23 may create an answer to a question indicated by question data based only on sensing data. In this case, the answer creation unit 23 also provides the question data to the learning model 41 and acquires, from the learning model 41, an answer template for the question indicated by the question data. For example, if the question indicated by the question data is, "Is Mr. / Ms. X sleeping now?", the answer creation unit 23 acquires, for example, an answer template (1) of "Mr. / Ms. X sleeping now" and an answer template (2) of "Mr. / Ms. X not sleeping now." If the sensing data from the bed sensor 61 or the sensing data from the room sensor 62 indicates that Mr. / Ms. X is sleeping, the answer creation unit 23 outputs the answer template (1) to the answer data return unit 24 as answer data indicating an answer corresponding to the question. If the sensing data of the bed sensor 61 or the sensing data of the room sensor 62 indicates that Mr. / Ms. XX is awake, the answer creation unit 23 outputs an answer template (2) to the answer data return unit 24 as answer data indicating the answer corresponding to the question.

[0041] For example, if the question indicated by the question data is "Is X-san eating?", then, for example, an answer template "X-san is eating" is acquired as answer template (1) and an answer template "X-san is not eating" is acquired as answer template (2). If the sensing data of the ingestion detection sensor 66 indicates that X-san is eating, the answer creating unit 23 outputs answer template (1) to the answer data replying unit 24 as answer data indicating the answer corresponding to the question. If the sensing data of the ingestion detection sensor 66 indicates that X-san is not eating, the answer creating unit 23 outputs answer template (2) to the answer data replying unit 24 as answer data indicating the answer corresponding to the question.

[0042] The answer data reply unit 24 acquires the answer data from the answer creation unit 23. The answer data reply unit 24 replies with the answer data to the chatbot engine unit 13 of the chatbot 1 (step ST4 in FIG. 4).

[0043] The chatbot engine unit 13 acquires answer data from the answer data reply unit 24. As shown in FIG. 5, the chatbot engine unit 13 displays the answer indicated by the answer data on the display of the display unit 14. The chatbot engine unit 13 also outputs the answer indicated by the answer data aloud from the speaker of the audio output unit 15. FIG. 5 is an explanatory diagram showing the answer "You need ◇◇ ml more" when the question is "How much more water does Mr. / Ms. XX need?". Note that ◇◇ contains the difference between "The amount of water Mr. / Ms. XX needs to drink per day" and "The amount of water Mr. / Ms. XX has drunk today." In FIG. 5, □□□□ is the name of the staff member.

[0044] In the above-described first embodiment, the question answering apparatus 2 is configured to include a sensing data collection unit 21 that collects sensing data from sensors that monitor the condition of the care recipient, and a question data acquisition unit 22 that acquires question data indicating a question about the care recipient. The question answering apparatus 2 also includes an answer creation unit 23 that creates an answer to a question indicated by the question data acquired by the question data acquisition unit 22, based on the sensing data collected by the sensing data collection unit 21, and an answer data reply unit 24 that replies with answer data indicating the answer created by the answer creation unit 23. Therefore, the question answering apparatus 2 can create an answer to a question that reflects the condition of the care recipient.

[0045] 1 , answer creation unit 23 provides question data to learning model 41 and acquires a template of an answer to the question indicated by the question data from learning model 41. However, this is merely an example, and answer creation unit 23 may acquire a template of an answer to the question indicated by the question data based on a rule base, for example.

[0046] In the question answering device 2 shown in FIG. 1 , a question includes information indicating that the care recipient is, for example, “Mr. / Ms. XX” or “Mr. / Ms. △△.” However, this is merely an example, and the question does not necessarily include information indicating that the care recipient is, for example, “Mr. / Ms. XX” or “Mr. / Ms. △△.” In this case, for example, when a staff member at a nursing home or the like speaks to the voice recognition unit 11 of the chatbot 1, the staff member provides the chatbot engine unit 13 with their ID (IDentification). The chatbot engine unit 13 then displays information indicating one or more care recipients for which the staff member is responsible on the display unit 14. The chatbot engine unit 13 then accepts a selection operation by the staff member to select a care recipient related to the question from one or more care recipients, and outputs care recipient information indicating the care recipient related to the question to the question answering device 2. The sensing data collection unit 21 collects sensing data related to the care recipient indicated in the care recipient information from the sensing data recorded in the sensing data recording unit 67 of the sensing device 3.

[0047] Second Embodiment In a second embodiment, a question and answering device 2 in which the answer creating unit 25 creates advice corresponding to the condition of the care recipient in addition to an answer to a question will be described.

[0048] Fig. 6 is a configuration diagram showing a system including a question answering apparatus 2 according to embodiment 2. In Fig. 6, the same reference numerals as in Fig. 1 indicate the same or corresponding parts, and detailed description thereof will be omitted. Fig. 7 is a hardware configuration diagram showing the hardware of the question answering apparatus 2 according to embodiment 2. In Fig. 7, the same reference numerals as in Fig. 2 indicate the same or corresponding parts, and detailed description thereof will be omitted. The question answering apparatus 2 includes a sensing data collection unit 21, a question data acquisition unit 22, an answer creation unit 25, and an answer data reply unit 26.

[0049] The answer creation unit 25 is realized, for example, by an answer creation circuit 35 shown in FIG. 7 . The answer creation unit 25 acquires sensing data from the sensing data collection unit 21 and acquires question data from the question data acquisition unit 22. The answer creation unit 25 creates an answer to a question indicated by the question data based on the sensing data and the care record of the care recipient. The answer creation unit 25 also creates advice corresponding to the condition of the care recipient. The answer creation unit 25 outputs answer data indicating the created answer and advice data indicating the advice to the answer data reply unit 26.

[0050] The answer data reply unit 26 is realized, for example, by the answer data reply circuit 36 ​​shown in Fig. 7. The answer data reply unit 26 acquires answer data and advice data from the answer creation unit 25. The answer data reply unit 26 replies to the answer data and advice data to the chatbot engine unit 13 of the chatbot 1.

[0051] 6, it is assumed that each of the components of the question and answering apparatus 2, namely, the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 25, and the answer data reply unit 26, is realized by dedicated hardware such as that shown in Fig. 7. That is, it is assumed that the question and answering apparatus 2 is realized by a sensing data collection circuit 31, a question data acquisition circuit 32, an answer creation circuit 35, and an answer data reply circuit 36. Each of the sensing data collection circuit 31, the question data acquisition circuit 32, the answer creation circuit 35, and the answer data reply circuit 36 ​​corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination of these.

[0052] The components of the question answering apparatus 2 are not limited to those realized by dedicated hardware, and the question answering apparatus 2 may be realized by software, firmware, or a combination of software and firmware. When the question answering apparatus 2 is realized by software, firmware, or the like, programs for causing a computer to execute the processing procedures of the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 25, and the answer data reply unit 26 are stored in a memory 51 shown in Fig. 3. Then, a processor 52 shown in Fig. 3 executes the programs stored in the memory 51.

[0053] 7 shows an example in which each of the components of the question answering apparatus 2 is realized by dedicated hardware, while Fig. 3 shows an example in which the question answering apparatus 2 is realized by software, firmware, etc. However, this is merely an example, and some of the components of the question answering apparatus 2 may be realized by dedicated hardware, and the remaining components may be realized by software, firmware, etc.

[0054] Next, the operation of the question answering apparatus 2 shown in Fig. 6 will be described. However, since the components other than the answer creating unit 25 and the answer data replying unit 26 are the same as those of the question answering apparatus 2 shown in Fig. 1, the operation of the answer creating unit 25 and the answer data replying unit 26 will mainly be described here.

[0055] The answer creation unit 25 acquires sensing data from the sensing data collection unit 21 and acquires question data from the question data acquisition unit 22. The answer creation unit 25 also acquires the care record of the care recipient from the care record database 42. Similar to the answer creation unit 23 shown in FIG. 1 , the answer creation unit 25 creates an answer to a question indicated by the question data based on the sensing data and the care record of the care recipient. Similar to the answer creation unit 23 shown in FIG. 1 , the answer creation unit 25 may create an answer to a question indicated by the question data based only on the sensing data. Unlike the answer creation unit 23 shown in FIG. 1 , the answer creation unit 25 creates advice corresponding to the condition of the care recipient. The answer creation unit 25 outputs answer data indicating the answer and advice data indicating the advice to the answer data reply unit 26.

[0056] Specifically, if answer templates and advice are linked in the learning model 41, the answer creation unit 25 acquires advice corresponding to the answer template when acquiring the answer template from the learning model 41. If the answer creation unit 25 acquires an answer template such as "Mr. / Ms. XX needs ◇◇ ml more" from the learning model 41, the answer creation unit 25 acquires, for example, an advice such as "Mr. / Ms. XX has diabetes, so he / she has a target amount of water to drink. Please encourage him / her to drink more." If the answer creation unit 25 acquires an answer template such as "Mr. / Ms. XX has diabetes, so he / she has a target amount of water to drink. Please encourage him / her to drink more." from the learning model 41, the answer creation unit 25 acquires, for example, an advice such as "Mr. / Ms. XX has diabetes, so he / she has a target amount of water to drink. Please encourage him / her to drink more." In this case, the learning model 41 is given question data, template data, and advice data during learning, and learns the answer template and advice. The template data and advice data are each training data. During inference, when question data is given from the answer creation unit 23, the learning model 41 outputs template data indicating a template of an answer corresponding to the question indicated by the question data and advice data indicating advice.

[0057] 6, answer templates and advice are linked in learning model 41. However, this is merely an example, and question answering device 2 may be provided with an advice database that records advice corresponding to answer templates, and answer creating unit 25 may obtain advice corresponding to the answer template obtained from learning model 41 from the advice database.

[0058] When creating an answer to a question, the answer creation unit 25 may create a graph showing the change over time in the sensing data collected by the sensing data collection unit 21, or a table showing the change over time in the sensing data. For example, if the question is "How much more water does Mr. / Ms. XX need to drink?", the answer creation unit 25 creates a graph showing the change over time in the sensing data of the ingestion detection sensor 66 that is recorded in the sensing data recording unit 67, or a table showing the change over time in the sensing data of the ingestion detection sensor 66. In this case, the sensing data of the ingestion detection sensor 66 is recorded in the sensing data recording unit 67 as time-series data. If the answer creation unit 25 creates a graph showing the change over time in the sensing data or a table showing the change over time in the sensing data, the answer creation unit 25 outputs the graph or the like to the answer data reply unit 26.

[0059] The answer data reply unit 26 acquires answer data and advice data from the answer creation unit 25. The answer data reply unit 26 replies to the answer data and advice data to the chatbot engine unit 13. If the answer data reply unit 26 acquires a graph or the like from the answer creation unit 25, it replies to the chatbot engine unit 13 with the graph or the like.

[0060] The chatbot engine unit 13 acquires answer data and advice data from the answer data reply unit 26. The chatbot engine unit 13 displays the answer indicated by the answer data and the advice data indicated by the advice data on the display of the display unit 14. If the chatbot engine unit 13 acquires a graph or the like from the answer data reply unit 26, it displays the graph or the like on the display of the display unit 14. The chatbot engine unit 13 also outputs the answer indicated by the answer data and the advice data indicated by the advice data from the speaker of the audio output unit 15. Figure 8 is an explanatory diagram showing the answer "You need ◇◇ ml more" and the advice "Because Mr. / Ms. XX has diabetes, he / she has a target amount of water to drink. Please encourage him / her to drink as much as possible" when the question is "How much more water does Mr / Ms. XX need?", along with a graph showing changes in sensing data over time.

[0061] In the above-described second embodiment, question answering device 2 is configured so that answer creating unit 25 creates advice corresponding to the situation of the care recipient in addition to an answer to a question indicated by the question data acquired by question data acquiring unit 22, and answer data returning unit 26 returns advice data indicating the advice created by answer creating unit 25 in addition to answer data indicating the answer created by answer creating unit 25. Therefore, question answering device 2 can create answers that reflect the situation of the care recipient as answers to questions, and can also create advice that corresponds to the situation of the care recipient.

[0062] Embodiment 3 In Embodiment 3, a question and answering device 2 will be described in which the answer creating unit 27 acquires, as the care record of the care recipient, information indicating the medical history of the care recipient, information about the care recipient recorded on a face sheet, or information indicating the schedule of the care recipient.

[0063] Fig. 9 is a configuration diagram showing a system including a question answering apparatus 2 according to embodiment 3. In Fig. 9, the same reference numerals as in Figs. 1 and 6 indicate the same or corresponding parts, and detailed description thereof will be omitted. Fig. 10 is a hardware configuration diagram showing the hardware of the question answering apparatus 2 according to embodiment 3. In Fig. 10, the same reference numerals as in Figs. 2 and 7 indicate the same or corresponding parts, and detailed description thereof will be omitted. The question answering apparatus 2 includes a sensing data collection unit 21, a question data acquisition unit 22, an answer creation unit 27, and an answer data reply unit 24.

[0064] The care record database 43 stores the care records of the care recipient. The care record database 43 also stores, as the care record of the care recipient, information indicating the medical history of the care recipient, information about the care recipient recorded on a face sheet, or the care recipient's schedule. The information about the care recipient recorded on the face sheet includes, for example, the care recipient's name, age, address, family composition, hobbies, favorite things, or work history. The care recipient's schedule includes, for example, a schedule for visiting the hospital, a schedule for attending a hobby class, or a schedule for going to the bank. The care record database 43 also stores the care recipient's house call history or the care recipient's prescription record.

[0065] In the system shown in Fig. 9 , information indicating the medical history of the care recipient, information about the care recipient recorded on a face sheet, the care recipient's schedule, the care recipient's visit history, or the care recipient's medication record is stored in the care record database 43. However, this is merely an example, and the information indicating the medical history of the care recipient, information about the care recipient recorded on a face sheet, the care recipient's schedule, the care recipient's visit history, or the care recipient's medication record may be stored in a database different from the care record database 43. In the system shown in Fig. 9 , the learning model 41 and the care record database 43 are each provided outside the question answering apparatus 2. However, this is merely an example, and the learning model 41 and the care record database 43 may each be provided inside the question answering apparatus 2.

[0066] The answer creation unit 27 is realized by, for example, an answer creation circuit 37 shown in FIG. 10 . The answer creation unit 27 acquires sensing data from the sensing data collection unit 21 and acquires question data from the question data acquisition unit 22. The answer creation unit 27 creates an answer to a question indicated by the question data based on the sensing data and the care record of the care recipient. When the answer creation unit 27 acquires the care recipient's house call history or the care recipient's prescription record, the answer creation unit 27 creates an answer to the question indicated by the question data based on the sensing data, the care record of the care recipient, and the house call history or the prescription record of the care recipient. The answer creation unit 27 outputs answer data indicating the created answer to the answer data reply unit 24.

[0067] 9 assumes that each of the components of the question and answering apparatus 2, namely, the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 27, and the answer data reply unit 24, is realized by dedicated hardware such as that shown in Fig. 10. That is, it assumes that the question and answering apparatus 2 is realized by a sensing data collection circuit 31, a question data acquisition circuit 32, an answer creation circuit 37, and an answer data reply circuit 34. Each of the sensing data collection circuit 31, the question data acquisition circuit 32, the answer creation circuit 37, and the answer data reply circuit 34 corresponds to, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC, an FPGA, or a combination of these.

[0068] The components of the question answering apparatus 2 are not limited to those realized by dedicated hardware, and the question answering apparatus 2 may be realized by software, firmware, or a combination of software and firmware. When the question answering apparatus 2 is realized by software, firmware, or the like, programs for causing a computer to execute the processing procedures of the sensing data collection unit 21, the question data acquisition unit 22, the answer creation unit 27, and the answer data reply unit 24 are stored in a memory 51 shown in Fig. 3. Then, a processor 52 shown in Fig. 3 executes the programs stored in the memory 51.

[0069] 10 shows an example in which each of the components of the question answering apparatus 2 is realized by dedicated hardware, while Fig. 3 shows an example in which the question answering apparatus 2 is realized by software, firmware, etc. However, this is merely an example, and some of the components of the question answering apparatus 2 may be realized by dedicated hardware, and the remaining components may be realized by software, firmware, etc.

[0070] Next, the operation of the question answering apparatus 2 shown in Fig. 9 will be described. However, since the components other than the answer creating unit 27 are the same as those of the question answering apparatus 2 shown in Fig. 1, the operation of the answer creating unit 27 will mainly be described here.

[0071] The answer creation unit 27 provides the question data to the learning model 41 and acquires, from the learning model 41, answer templates for the question indicated by the question data. If the question indicated by the question data is, for example, "Has there been any change in Mr. / Ms. XX recently?", the answer creation unit 27 acquires answer templates such as those shown below from the learning model 41. For example, the answer creation unit 27 acquires an answer template of "There has been no change in Mr. / Ms. XX" as answer template (1), an answer template of "I have noticed a change in Mr. / Ms. XX. He is sleeping less" as answer template (2), an answer template of "I have noticed a change in Mr. / Ms. XX. He is eating less" as answer template (3), and an answer template of "I have noticed a change in Mr. / Ms. XX. He is spending less time on his hobbies" as answer template (4).

[0072] The answer creation unit 27 acquires information on "X's average sleeping hours," information on "X's daily food needs," or information on X's hobbies recorded on the face sheet from X's care records stored in the care record database 43. The answer creation unit 23 acquires information on "X's sleeping hours yesterday" from the sensing data of the bed sensor 61 or the sensing data of the room sensor 62 recorded in the sensing data recording unit 67. The answer creation unit 23 acquires information on "X's food intake yesterday" from the sensing data of the intake detection sensor 66 recorded in the sensing data recording unit 67. The answer creation unit 23 acquires information on the amount of time X spent on hobbies from X's care records.

[0073] The answer creation unit 23 compares "X's average sleep time" with "X's sleep time yesterday," and compares "X's daily food intake required" with "X's food intake yesterday." If X's sleep time yesterday is less than X's average sleep time, the answer creation unit 23 selects answer template (2) from among answer templates (1) to (4). If X's food intake yesterday is less than X's daily food intake required, the answer creation unit 23 selects answer template (3) from among answer templates (1) to (4). If X's time spent on hobbies is on a decreasing trend, the answer creation unit 23 selects answer template (4) from among answer templates (1) to (4). The answer creation unit 23 selects answer template (1) from answer templates (1) to (4) if XX's sleep time yesterday is equal to or longer than XX's average sleep time, XX's sleep time yesterday is equal to or longer than XX's average sleep time, and XX's time spent on hobbies is not on a decreasing trend. The answer creation unit 23 outputs the selected answer template as the created answer to the answer data reply unit 24, and answer data indicating the answer.

[0074] Here, the answer creation unit 27 acquires answer templates (1) to (4) from the learning model 41. The answer creation unit 27 may acquire answer template (5) from the learning model 41. An example of answer template (5) is a template such as "There is a change in Mr. / Ms. X. X.'s schedule is busy." The answer creation unit 23 acquires information about "Ms. X.'s schedule" from the care record of Mr. / Ms. X. and compares how busy Mr. / Ms. X.'s schedule is with the average schedule of people of the same age as Mr. / Ms. X. If the schedule of Mr. / Ms. X. is busier than the average schedule of people of the same age as Mr. / Ms. X., the answer creation unit 23 selects answer template (5) from answer templates (1) to (5).

[0075] Here, the answer creation unit 27 acquires answer templates (1) to (5) from the learning model 41. The answer creation unit 27 may acquire answer template (6) from the learning model 41. An example of answer template (6) is a template such as, "There is a change in Mr. / Ms. X. There may be something abnormal with Mr. / Ms. X." The answer creation unit 23 acquires information on "Mr. / Ms. X's house call history" from Mr. / Ms. X's care record. If the care record includes a house call history for Mr. / Ms. X within the past week, for example, the answer creation unit 23 selects answer template (6) from answer templates (1) to (6).

[0076] Here, the answer creation unit 27 acquires answer templates (1) to (6) from the learning model 41. The answer creation unit 27 may acquire answer template (7) from the learning model 41. An example of answer template (7) is a template such as "A change is being noticed in Mr. / Ms. XX. The types of medications Mr. / Ms. XX takes or the amount of medications Mr. / Ms. XX takes has increased." The answer creation unit 23 acquires information on "Mr. / Ms. XX's prescription record" from Mr. / Ms. XX's nursing care record. If the types of medications Mr. / Ms. XX takes have increased or the amount of medications Mr. / Ms. XX takes has increased, the answer creation unit 23 selects answer template (7) from answer templates (1) to (7).

[0077] The question answering device 2 according to the third embodiment can create answers to questions that better reflect the condition of the care recipient than the question answering device 2 according to the first embodiment.

[0078] Embodiment 4. In the question answering device 2 according to Embodiments 1 to 3, the answer creation units 23, 25, and 27 create answers to questions indicated by question data acquired by the question data acquisition unit 22 based on sensing data collected by the sensing data collection unit 21. In addition to creating answers to questions, the answer creation units 23, 25, and 27 may determine the degree of independence of the care recipient based on the usage status of a home appliance, such as a refrigerator, indicated by monitoring results from a home appliance sensor 63 attached to the home appliance. Specifically, if the home appliance to which the home appliance sensor 63 is attached is, for example, a refrigerator, the answer creation units 23, 25, and 27 calculate the frequency with which the refrigerator door is opened and closed based on the monitoring results from the home appliance sensor 63. If the home appliance to which the home appliance sensor 63 is attached is, for example, a microwave oven, the answer creation units 23, 25, and 27 calculate the frequency with which the microwave oven is used based on the monitoring results from the home appliance sensor 63. These frequencies may be, for example, daily, weekly, or monthly. The answer creation units 23, 25, 27 determine, for example, that the higher the frequency, the higher the degree of independence of the care recipient. The higher the frequency, the higher the degree of independence, and it does not matter whether the relationship between the frequency and the degree of independence is linear or not. The answer data reply units 24, 26 output the independence degree determination results made by the answer creation units 23, 25, 27 to the chatbot engine unit 13. The chatbot engine unit 13 acquires the independence degree determination results from the answer data reply units 24, 26. The chatbot engine unit 13 displays the independence degree determination results on the display unit 14, or outputs the independence degree determination results by voice from the voice output unit 15.

[0079] In addition, the present disclosure allows for free combination of the respective embodiments, modification of any of the components of the respective embodiments, or omission of any of the components of the respective embodiments.

[0080] The present disclosure is suitable for a question and answering device and a question and answering method.

[0081] 1 Chatbot, 2 Question and answer device, 3 Sensing device, 11 Speech recognition unit, 12 Speech database, 13 Chatbot engine unit, 14 Display unit, 15 Speech output unit, 21 Sensing data collection unit, 22 Question data acquisition unit, 23 Answer creation unit, 24 Answer data reply unit, 25 Answer creation unit, 26 Answer data reply unit, 27 Answer creation unit, 31 Sensing data collection circuit, 32 Question data acquisition circuit, 33 Answer creation circuit, 34 Answer data reply circuit, 35 Answer creation circuit, 36 Answer data reply circuit, 37 Answer creation circuit, 41 Learning model, 42, 43 Care record database, 51 Memory, 52 Processor, 61 Bed sensor, 62 Room sensor, 63 Home appliance sensor, 64 Human presence sensor, 65 Vital sign sensor, 66 Intake detection sensor, 67 Sensing data recording unit.

Claims

1. A question answering device comprising: a sensing data collection unit which collects sensing data from a sensor which monitors the condition of a care recipient; a question data acquisition unit which acquires question data indicating a question regarding the care recipient; an answer creation unit which creates an answer to a question indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit; and an answer data reply unit which replies with answer data indicating the answer created by the answer creation unit.

2. The question answering device described in claim 1, characterized in that the answer creation unit creates an answer to the question indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit and the care records of the care recipient.

3. The question answering device described in claim 1, characterized in that the answer creation unit creates advice corresponding to the situation of the care recipient in addition to an answer to the question indicated by the question data acquired by the question data acquisition unit, and the answer data reply unit replies with advice data indicating the advice created by the answer creation unit in addition to answer data indicating the answer created by the answer creation unit.

4. The question answering device described in claim 1, characterized in that the answer creation unit creates, in addition to the answer to the question, a graph showing the change over time in the sensing data collected by the sensing data collection unit, or a table showing the change over time in the sensing data, and the answer data reply unit replies with the graph or the table in addition to the answer data showing the answer created by the answer creation unit.

5. The question answering device described in claim 2, characterized in that the answer creation unit acquires, as the care record of the care recipient, information indicating the medical history of the care recipient, information about the care recipient recorded on a face sheet, or information indicating the schedule of the care recipient, and creates an answer to a question indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit and the care record of the care recipient.

6. The question and answering device described in claim 2, characterized in that the answer creation unit creates answers to questions indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit and the care records of the care recipient, as well as the house call history of the care recipient.

7. The question and answering device described in claim 2, characterized in that the answer creation unit creates answers to questions indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit and the care records of the care recipient, as well as the prescription records of the care recipient.

8. The question answering device described in claim 1, characterized in that the sensor is a home appliance sensor that monitors the usage status of home appliances used by the care recipient, and the answer creation unit not only creates answers to questions indicated by the question data acquired by the question data acquisition unit, but also determines the degree of independence of the care recipient based on the usage status of the home appliances indicated by the monitoring results of the home appliance sensor.

9. A question answering method, in which a sensing data collection unit collects sensing data from a sensor monitoring the condition of a care recipient, a question data acquisition unit acquires question data indicating a question about the care recipient, an answer creation unit creates an answer to the question indicated by the question data acquired by the question data acquisition unit based on the sensing data collected by the sensing data collection unit, and an answer data reply unit replies with answer data indicating the answer created by the answer creation unit.

Images